dharma-incident-commander

Use an incident command structure that assigns severity first, then establishes containment, maps owners to each workstream, and keeps a running timeline. This Skill enforces exactly that sequence and sets a next update time for stakeholder communication.

An incident commander should produce five artifacts: a severity assignment, a containment plan, an owner map, an incident timeline, and a scheduled next update time. These outputs keep response coordinated and stakeholders informed.

Use it when there is active breakage, customer impact, a security concern, or uncertainty about severity. Incidents are not normal feature work, so they require explicit command structure rather than parallel unstructured effort.

Yes, the rules explicitly require stating clearly when rollback is the right move. The containment planning step surfaces rollback as an option rather than hiding it behind continued debugging.

The workflow hands off to engineering review, then QA, and finally a retrospective. This ensures the incident leads to verified fixes and organizational learning rather than ending at containment.